Aji Suraji, Candra Aditya, Prabowo Prabowo, Muhammad Fatkhurrozi, Gholiqul Amrodh Alawy, Rangga Pahlevi Putra
Road condition survey is a job that requires precision, accuracy, and speed in field implementation and analysis. This study aims to identify cracking classification on rigid pavement using the Machine Learning Method, namely Multiclass Support Vector Machine (MSVM). The survey method used is to collect data on the condition of rigid pavement with a direct survey using a photo camera. The analysis method uses MSVM, where the images that have been obtained are processed and analyzed by looking at the value of energy features and kernel type accuracy. In the analysis stage of the value of energy features, polynomials and Gaussians are used. The results of this study show that the highest energy feature value in the type of crack damage is evenly distributed with a value between 3.46 to 3.60. Meanwhile, kernel type accuracy analysis with a polynomial approach in general has a higher tendency than Gaussian. The results of this study show that the polynomial approach has better accuracy, and it is recommended for researchers to use this approach. © 2025 IEEE.
University of Widya Gama Malang, Dept. of Civil Engineering, Malang City, Indonesia; Universitas Negeri Malang, Dept. of Eng. Tech and Maintenance of Civil Buildings, Malang City, Indonesia; Institut Teknologi Sumatera Lampung Selatan Regency, Dept. of Ocean Engineering, Indonesia; Jember University, Dept. of Civil Engineering, Jember, Indonesia
